the contribution scaling the catalanprosodia.upf.edu/home/arxiu/tesis/master/tesina_nadeu.pdfblondet...
TRANSCRIPT
THE CONTRIBUTION OF TONAL SCALING TO
THE PERCEPTION OF POLITENESS IN CATALAN
Marianna Nadeu
Supervisor: Pilar Prieto
MA in Cognitive Science and LanguageResearch paperDecember 2008
ACKNOWLEDGMENTS
I would like to extend my heartfelt thanks to everyone who contributed to this
research paper in one way or another. First and foremost, I am especially
grateful to Professor Pilar Prieto for introducing me to the study of intonation
and agreeing to be my advisor, as well as for her guidance, suggestions,
approachability and support in conducting this piece of research and
elsewhere. I would like to thank Teresa Cabré and Eulàlia Bonet for being part
of my evaluation committee.
I also want to thank Professor José Ignacio Hualde for reading a draft of the
paper and providing me with insightful comments. Thanks also go to Professor
Marina Terkourafi for suggesting readings and allowing me to audit her class.
A very special thank you is also due to Professor Joan Carles Mora for sparking
my interest in research in the first place, and also for technical assistance
preparing the test, conducting some of the tests and for reading a preliminary
version. I am especially thankful to Maria del Mar Vanrell for her invaluable
help in technical matters, her friendliness and encouragement, especially
when I was most lost. I also want to thank Núria Argemí for lending me her
voice when I lost mine.
Finally, I am really indebted to all the people who participated in the
experiment. Needless to say, this paper would have been impossible without
their contribution. I also want to extend a warm hug to my family and friends,
for simply being there and encouraging me whenever I need it.
2
ABSTRACT
It has been proposed in the literature that prosody plays a role in the
expression of politeness. The present paper attempts to explore the
contribution of tonal scaling to the expression of paralinguistic meanings.
More specifically, it examines the relationship between pitch height and the
degree of perceived politeness in yes–no questions in Central Catalan.
Studies on the biologically–determined use of pitch (Gussenhoven, 2002;
Nilsenova, 2006; Ohala, 1984, 1996) claim that high or rising F0 can
convey politeness. Experimental studies have found that use of a higher
pitch range or use of higher peaks is typical of polite utterances (Álvarez &
Blondet, 2003; Maekawa, 1999; Orozco, 2006). Yet, other studies
(Culpeper, 2005) have shown that higher pitch can also be associated to
impoliteness. In order to test whether pitch height correlates positively or
negatively with politeness, two continua were created by gradually
increasing and decreasing the final boundary tone of a rising and a falling
yes–no question. 20 native speakers of Central Catalan took part in the
experiment. They had to compare stimuli in pairs and also rate the degree
of politeness that each stimulus conveyed in isolation. Results obtained in
both tasks and for both continua point to the same direction: the degree of
perceived perception gradually decreases when the pitch height of the final
boundary tone is increased.
3
1. INTRODUCTION
The study of the linguistic expression of politeness has traditionally focused on
lexical choice and on morphosyntactic aspects of the utterance, such as the
use of indirect structures, formal forms of address, or conventionalized
politeness markers (see Watts, 2003, for a classification of expressions and a
discussion of their value). Even if this is a less explored area, it is widely
acknowledged that prosody (intonation or use of pitch1, rate of speech, voice
quality, among other elements) contributes to the perception of politeness in
different languages. Although they do not dwell much in it, Brown and Levinson
(1987) list exaggeration (of interest, approval, sympathy with the hearer) as a
positive politeness strategy which may be implemented by means of
“exaggerated intonation, stress, and other aspects of prosodics” (p. 104). It is
unclear, though, what “exaggerated intonation” actually consists in. More
broadly, Wichmann and Cauldwell (2003) refer to the belief that emotions can
be revealed by certain acoustic parameters, such as pitch, loudness and
speed, either globally (stretching over the utterance) or locally. Chen (2005)
talks about the paralinguistic meaning of intonation, i.e. degrees of a given
meaning can be conveyed by gradual changes in pitch range or peak
alignment. Crucially, this type of meaning is gradient, not categorical.
Research conducted on the contribution of prosody to the perception of
politeness is not very extensive. Studies so far have focused on peak
alignment (Maekawa, 1999), choice of nuclear configuration type (Estrella–
Santos, 2007; Payà, 2003), F0 movements (final rising or falling boundary
tone), rate of speech (Ofuka, McKeown, Waterman, and Roach, 2000; Payà, 1 Throughout the paper, terms such as "pitch", "pitch height", “pitch scaling”, and "pitch range" will be used. A brief definition is provided here for those concepts. Rate of vibration of the vocal folds determines pitch, "the auditory sensation of tonal height" (Gussenhoven, 2004:1). Higher frequency of vibration of the vocal folds (higher F0 or fundamental frequency) results in higher pitch, and it is expressed in hertz (Hz). "Pitch height" is, then, determined by vocal fold vibration. “Pitch scaling” refers to changes in pitch height. "Pitch range" refers to the distance or span between the lowest and the highest F0 values (a valley and a peak, respectively) observed in an utterance. Pitch range is automatically modified when pitch height increases or decreases at one point in the utterance but remains unchanged in the rest. None of the studies reviewed in this paper manipulate "register". By varying register, a whole contour is raised or lowered in the F0 space. This will not be examined here. For more information on these concepts, see Hualde, 2005; Gussenhoven, 2004; Prieto, 2002a, 2002b.
4
2003), pitch range (Orozco, 2006), and pitch height together with F0 variability
(Álvarez and Blondet, 2003). Some studies assess the contribution of prosody
vis–à–vis lexical choice or morphological configuration (Estrella–Santos,
2007; Maekawa, 1999; Payà, 2003).
The relationship between prosodic aspects and politeness in Catalan has been
studied by Payà (2003). Specifically, she examined the contribution of choice
of pitch contour type and speed of delivery to the perception of politeness. She
analyzed the interpretation of utterances with conflicting prosodic and non–
prosodic elements (the former conveying politeness but not the latter, or vice
versa). By so doing, she wanted to isolate prosodic strategies and assess their
contribution to the perception of politeness. In her study, she used the two
common contours for yes–no questions in Catalan: the falling contour as the
most polite, and the rising one as the least polite. As regards lexical and
morphological choice, one of the utterances was clearly impolite (Vols venir
d’una puta vegada? ‘Do you want to fucking come?’), while the other was
clearly polite (Que li faria res venir un momentet? ‘Would you–formal mind
coming for a second?’). In her stimuli, choice of pitch contour entailed some
variation in duration and intensity as well. The results showed that the
appearance of any impolite cue (whether prosodic or non–prosodic) overrode
the rest and, thus, biased the hearer to an impolite interpretation.
Other studies have explored the effects of intonation on politeness in another
Romance language, Spanish, through both production and perception studies.
In Álvarez and Blondet (2003), two Venezuelan Spanish interrogatives were
compared: an informational wh–question and a “polite sentence” (indirect
command). Sixteen utterances were recorded (two interrogatives repeated
once by eight women) and measures were taken for each syllable in them. The
comparison revealed that the polite yes–no question displayed more F0
variability (more peaks and valleys), whereas the wh–question was rather flat.
Similarly, the authors argued that there was a difference in pitch height: the
polite yes–no question being higher than the wh–question. Álvarez and
Blondet (2003) hypothesize that differences would be found in the length of
syllables too, so other prosodic elements may be having an effect. In fact, the
two interrogatives examined were not directly compared because they were of
a different type: the polite sentence was a yes–no question (¿Me compras el
5
periódico? ‘Can you buy me the newspaper?’), whereas the question was a
wh–question (¿Cómo te fue el paseo? ‘How was your walk?’). In addition, the
former contained eight syllables, while the latter was made up of nine
syllables. Differences observed between the two interrogatives might be due to
the linguistic material in them.
Orozco (2006) studied the prosodic realization of interrogative requests in
Mexican Spanish. She asked participants to read three sentences2 (printed
without punctuation signs) in three different ways: in a “normal”, “dry”, and
“kind” way (p. 6). She compared the neutral vs. polite realizations by gathering
F0 measurements at several points: initial (first syllable) and final boundary
tone; beginning, mid point and end of the stressed syllable; as well as mid
points of the pretonic and the posttonic syllables. As far as pitch contour is
concerned, no differences between neutral vs. polite utterances were
observed. However, polite requests tended to begin with a %H boundary tone
(neutral ones were more often realized with a %L boundary tone). Another
difference observed concerned pitch range: polite expressions were produced
with an expanded pitch range. This difference became more evident when
analyzing women’s speech separately: women used a wider pitch range than
men in their polite requests.
In a perception study, Estrella–Santos (2007) addressed the relationship
between pitch contour type and politeness in a variety of Spanish in contact
with Quechua. Estrella–Santos noted that Ecuadorian Andean Spanish
speakers used both morphological and prosodic cues to determine the degree
of politeness of requests. The variety of Spanish she studied exhibits a polite
request structure which is actually a calque of a Quechua structure containing
a benefactive verb (dar ‘to give’). Both the morphological and the prosodic
Quechua configuration made their way into Spanish. In her experiment, she
used Spanish imperatives (with their specific prosody and also with the
Quechua intonational configuration) and Quechua–borrowed request
structures (with their own prosody and with the Spanish imperatives’). When
2 The sentences were: Podrías apagar tu cigarro (‘Could put out your cigarette?’), Te molesta si fumo (‘Do you mind if I smoke?’), Échame una mano en la cocina (‘Help me out in the kitchen’). The first two sentences have been rendered as questions in English but, strictly speaking, since they did not include punctuation marks in Spanish, they could be interpreted as either interrogative or declarative sentences.
6
comparing the Quechua calque (e.g. Deme pasando la sal ‘Benefactive give +
passing me the salt’) to the Spanish imperative (e.g. Páseme la sal ‘Pass me
the salt’), Ecuadorian speakers preferred the former over the latter in all
cases. When intonation was held constant (that of the Spanish imperative), the
Quechua structure with the benefactive verb was always preferred over the
Spanish imperative. When the Quechua intonation was superimposed on both
types of morphological structures, the Quechua structure was preferred over
the latter (69% vs. 31%). Finally, when structures exchanged their intonation
(the Spanish imperative was presented with the Quechua intonation and the
Quechua structure with the Spanish intonation), intonation seemed to yield
better results than morphology (53.5% vs. 46.5% respectively). In general,
then, the Quechua–borrowed intonational contour seems to play a crucial role
in the perception of politeness in Ecuadorian Andean Spanish.
In the Japanese culture, politeness expressed both linguistically and
attitudinally is considered of utmost importance (see, for example, Ide, 1982),
and its expression has been studied extensively. As regards the contribution of
prosody to it, Ofuka et al. (2000) explored the relationship between speech
rate and use of high vs. low pitch, and the perception of politeness. The
authors recorded several speakers interpreting a short scene with varying
interlocutors, which compelled them to use three different speaking styles:
“polite”, “casual” and “authoritative”. In the variety of Japanese investigated in
this study, pitch pattern is very constricted, allowing variation only at the end of
the sentence. The analyses of the utterances revealed that slower speech rate
and total duration of utterances correlated positively with politeness; however,
there was no consistency across speakers in the use of higher vs. lower pitch
to signal polite vs. casual speech. Only F0 movements at the end of the
utterance (but not F0 level, F0 range or F0 rate of change) were used
consistently to mark different speaking styles. The two features that were used
consistently in production (F0 movements at the end of the utterance and
speech rate) were further examined in two separate perception experiments.
The first perception test included three variables: speaking style (polite vs.
casual), final vowel duration (short vs. long), and F0 movement type (rise vs.
fall). Final vowel duration was found to play a crucial role (short vowels being
perceived as more polite), whereas F0 movement remained less influential.
Final rises were evaluated more positively than final falls, but the difference
7
failed to reach significance. The second perception test manipulated overall
speech rate and it was observed that politeness ratings depended on
participants’ own speech rate.
Maekawa (1999) examined the effect of pitch peak alignment and question
particle choice in another variety of Japanese, Kumamoto Japanese,
characterized by a ‘wandering H(igh)’ tone (any of the syllables in an accentual
phrase can receive the high pitch peak). In addition, Kumamoto Japanese
allows two different final boundary tones in questions, either falling or rising.
This freedom of alignment, as well as the different options of the final
boundary tone, results in a variety of contours which are equally grammatical,
but which vary as to the degree of politeness they convey. In a perception
experiment, Maekawa found that contours displaying a later peak (in the last
syllables of the utterance) or a rising boundary tone received higher politeness
scores than those with an early peak (toward the beginning of the utterance)
or ending in a falling cadence. The effects of peak alignment were statistically
significant, independently of the use of the particles compared. With this
study, Maekawa confirmed a model of the way in which politeness is affected
by intonation which he had proposed in previous works: the position of the last
intonational peak, together with its magnitude, determines the degree of
politeness expressed by the utterance. The later the peak and the larger its
prominence, the higher the degree of politeness perceived.
In a nutshell, the studies summarized above do explore the prosody–
politeness interface, and they tend to focus on intonational contour choice,
speech rate, pitch height, pitch range, pitch variation, and pitch peak
alignment. The same is true of studies on impoliteness.
Culpeper, Bousfield, and Wichmann (2003) analyzed impolite utterances
which aimed specifically at attacking the hearer’s face3. They observed that
prosody was involved in the expression of impoliteness in different ways: by
choosing a specific nuclear contour (which determines the illocutionary force
of the utterance), by allowing the conversation to continue or blocking it (by
3 It is important to understand that impoliteness is not simply absent or failed politeness. Impoliteness constitutes a deliberate attempt to damage the hearer’s face, according to Culpeper (2005) and Culpeper et al. (2003).
8
using rising or falling contours respectively), or by using extreme loudness or
pitch in order to invade the hearer’s auditory space. In a study of (theoretically
fake) impoliteness as entertainment (as exhibited in the quiz show The
weakest link), Culpeper (2005) confirmed that prosody plays an important role.
In his example, the expression of what is considered a polite farewell
(“goodbye”) is perceived as rude by virtue of its being accompanied by means
of “faster tempo, tense articulation and […] much higher pitch average” (p.
53). In addition, the intonation contour comprises a fall from a very high
starting point, which, he claims, indicates closure to the hearer and therefore
appears as impolite as the hearer is not given the choice to add anything else.
Some of the studies summarized above (Álvarez and Bondet, 2003; Culpeper,
2005; Culpeper et al., 2003, and Orozco, 2006) mention pitch height or range
as a means to convey a higher or lower degree of politeness. Interestingly, a
biological basis of the use of pitch variation has been posited in the literature.
Some authors have endorsed the biological nature of (at least, part of)
intonational meaning. Ohala (1984, 1996) suggested the existence of a
cross–species phylogenic use of F0, which he labeled the frequency code.
According to this view, the current use of pitch in human languages has
evolved from a primitive code which relates high or rising F0 and low or falling
F0 to basic meanings of “smallness” and “bigness” respectively and to
secondary social meanings of “subordination, submission, lack of threat or
confidence, politeness” and “dominance, threat, authority, aggression,
assertiveness” respectively. Nilsenova (2006) adds the question of age in
relation to the frequency code: higher pitch signals that the speaker is either
very young or very old, hinting again at a situation of unequal power relations.
Similarly, Culpeper et al. (2003) claim that the meanings conveyed by prosody
can be “iconic or physiological in origin”. They associate high pitch to “small,
young” whereas low pitch is linked to “big, old”. This dichotomy can be
extended such that high pitch is associated with deference (“behaving in a
‘small’ way”).
The present study aims at pursuing a line of research that has not received
much attention yet. The goal behind it is to explore the pragmatic contribution
of pitch scaling. Specifically, we seek to determine how differences in pitch
height affect the perception of politeness of interrogative requests in the
9
Central variety of Catalan in a perceptual experiment, given that some studies
have proposed that pitch can be used to convey nuances (other than
emphasis) such as politeness or submission. In fact, the study described here
is (to our knowledge) the only perception study where pitch height of the final
boundary tone of the interrogative stimuli is manipulated in order to assess its
contribution to the perception of politeness. Two different hypotheses can be
proposed based on the studies just reviewed. Pitch height of the final
boundary tone is expected to correlate positively with degree of politeness,
based on the claims sustained both in studies on the biological use of pitch
and also in some experimental studies reviewed above (Álvarez and Blondet,
2003; Maekawa, 1999; Orozco, 2006). Nevertheless, Culpeper’s (2005) study
associated higher pitch with impoliteness, so pitch height can also be
predicted to correlate negatively with perception of politeness according to
that article.
In order to test which of the two hypotheses holds true, 18 synthetic stimuli
were created by increasing and decreasing the final boundary tone of two yes–
no questions in equidistant steps. Each of the two base questions displayed
one of the two possible contours in Catalan: one of the questions ended in a
rising boundary tone, whereas the other one ended in a falling boundary tone.
Speakers of Central Catalan were asked to judge which of two modified stimuli
presented one after the other was more polite in a given situation in a
comparison task, as well as to rate the degree to which individual stimuli
seemed polite or appropriate to a specific situation on a five–point scale of
politeness. A fully–fledged description of the experiment can be found in
section 2.
The rest of the article is organized as follows. Section 2 describes the two
Catalan contours used in this paper and provides information on the
methodology. Section 3 presents and discusses the results obtained. Finally,
section 4 summarizes the conclusions.
10
2. METHODOLOGY
2.1 Goals of the studyAs has been exposed before, this paper deals with the prosody–politeness
(phonology–pragmatics) interface. Following studies on the biologically–
determined use of pitch and its signification (Gussenhoven, 2002; Nilsenova,
2006; Ohala, 1984, 1996; among others) and studies on the effects of
prosody on the perception of politeness (Álvarez and Blondet, 2003; Culpeper,
2005; Culpeper et al., 2003; Estrella–Santos, 2007; Maekawa, 1999; Ofuka
et al., 2000; Orozco, 2006; Payà, 2003), the research question is whether
pitch scaling modifies the degree of politeness perceived in both falling and
rising request interrogatives in Central Catalan. According to studies on the
biological codes, it is to be expected that higher final boundary tones elicit
higher politeness ratings, and that decrease of the final boundary tone be
perceived as a decrease in politeness. Following Álvarez and Blondet (2003)
and Orozco (2006), use of a higher pitch range (which is the result of
manipulating the final boundary tone while keeping the rest constant) is also
expected to correlate positively with politeness. However, studies on
impoliteness (Culpeper, 2005) have also shown that higher pitch can be
associated to impoliteness. Hence, another possibility is that higher pitch can
be associated with impoliteness, rather than politeness.
2.2 Catalan yes–no questionsThe use of interrogative sentences with the force of requests or commands
has been extensively noted in the literature on politeness and speech acts. In
her PhD dissertation, Escandell (1988) offered a comprehensive approach to
the syntax, semantics and pragmatics of Spanish interrogatives, of which she
also described the intonational contours. Escandell claims that “interrogation
functions, then, as a ‘syntactic euphemism’ to avoid the use of closed forms
[i.e. forms which do not let the hearer opt out] and the impoliteness
associated with them” (p. 582, my translation). In other words, interrogative
11
forms can be used whenever the speaker wants to compel the hearer to do
something without imposing on him/her. She assumes that interrogative
requests are “intrinsically polite” (p. 580) by virtue of letting the hearer accept
or deject the proposal. Escandell goes on to say that a polite request entails
the creation of a fictional situation in which the speaker assumes an inferior
position, thus conferring (fictional) authority to the hearer. By so doing, the
hearer does not consider the request an imposition, although, through his or
her knowledge of the world, s/he knows s/he does not have the authority
attributed to him/her and thus interprets the utterance as a polite request (p.
572–573). She then offers a summary of the conditions required for an
interrogative to be interpreted as a request (or what she calls an “impositive
directive interrogative”).
As Bonet (1984), Payà (2003), and Prieto (2001, 2002b) pointed out, yes–no
questions in Central Catalan allow two different intonational realizations: the
falling (figure 1) and rising (figure 2) contours. Prieto (2001) considers both
contours to be unmarked and describes them in the following terms. The
falling contour (figure 1) initiates at a relatively high pitch and maintains this
height up to the beginning of the last stressed syllable, where the pitch starts
to decline and reaches the speaker’s minimum F0 values. Typically, a slight
increase in pitch is observable in the syllable prior to the beginning of the
decline (Prieto, 2001, 2002a). In Central Catalan, this type of yes–no
questions may be introduced by the question particle que (Prieto, 2001; Prieto
and Rigau, 2007). Following the notation described for Cat_ToBI (Prieto,
Aguilar, Mascaró, Torres–Tamarit, and Vanrell, 2009; for an introduction in
English, see Prieto, in press), the falling pattern is transcribed as H+L* L%.
The other class of yes–no questions exhibits a rising contour (figure 2). This
contour initiates at a mid tone that decreases mildly over the first unstressed
syllables. A low pitch accent is anchored in the first stressed syllable. A rising
movement is associated with the following syllable, after which there is a
gradual pitch decrease up to the last stressed syllable, where another low
pitch accent is anchored. Finally, the contour concludes in a rising melody. This
contour, transcribed as L* HH%, is incompatible with the presence of que. For
a more comprehensive approach to interrogatives, see Prieto, 2001 (a
12
dialectological approach to interrogatives), 2002a, 2002b, in press; and Prieto
et al., 2009.
In Payà’s article (2003), the falling contour is described as the most polite.
However, Payrató (2002) claims that both patterns are adequate when the
speaker’s attitudes are neutral and when the aim of the interrogative is simply
to elicit unknown information. However, when the action conveyed in the
utterance implies a high benefit for the speaker and a high cost for the hearer,
the falling pattern is perceived as less polite or formal than the other. The
rising contour is preferred in those cases (Prieto, 2001; Payrató, 2002).
que te- niu man- da- ri- nes?
0
500
100
200
300
400
Pitc
h (H
z)
Time (s)0 1.011
Figure 1. Falling yes–no question (Que teniu mandarines? ‘Do you have tangerines?’). Note the falling movement (H+L*) starting in the pretonic “da–” and the final low
cadence (L%).
te- niu man- da- ri- nes?
0
500
100
200
300
400
Pitc
h (H
z)
Time (s)0 0.8805
Figure 2. Rising yes–no question (Teniu mandarines? ‘Do you have tangerines?’). Note the L* in the stressed syllable “ri–” and the HH% boundary tone.
13
2.3 Stimulus preparationTwo interrogative utterances, one with the falling and the other with the rising
contour, were uttered by a young female native speaker of Central Catalan.
They were recorded in a sound–proof booth. The content of the sentence was
Tens hora? (‘Do you have the time?’) and it was selected so that it remained
semantically neutral (following Ofuka et al., 2000): at least in a Catalan
environment, asking what time it is is a frequent situation and is not perceived
as imposing a high cost on the hearer. This was done in order to prevent the
form of the text from biasing participants’ responses. For the same reason, no
“politeness markers” (such as please) or polite forms of address were
included.
Both interrogatives were manipulated and synthesized using the PSOLA
technique in the Praat software (Boersma and Weenink, 2005). Each original
utterance was modified so that eight different new versions of each were
produced by increasing or decreasing the boundary tone by 6 Hz each time.
For the falling interrogative, the modification of the boundary tone affected the
pitch accent in order to avoid altering the contour. In the falling contour, the
boundary tone typically stays at the same level as the L* valley reached by the
pitch accent. Modifying only the boundary tone would result in a complex
contour which is not typical of Catalan falling yes–no questions. The total
number of stimuli was 18 (16 new modified ones and the two original ones,
which were also synthesized). Figures 3 and 4 show schematic
representations of the different stimuli created.
14
Figure 3. Falling yes–no question stimuli.
Figure 4. Rising yes–no question stimuli.
2.4 Test administration The test was administered by means of the DmDx Display Software (Forster,
2007). Participants were seated at a laptop wearing headphones. All the
instructions were printed on the screen in Catalan and participants were
allowed to read them at their own pace. The test consisted of four parts: two
comparison tasks and two rating tasks. Each task contained only rising or
15
falling stimuli. Rising and falling stimuli were not mixed in any task. In the
comparison tasks, participants were presented with two stimuli separated by a
1.2 second pause. They had to decide which of the two was the most polite by
pressing a key in the keyboard (“1” if the first stimulus was perceived as the
most polite; “2” if the second one was the most polite). Each pair was formed
by combining every other stimulus in the continuum (e.g. stimulus 1 and
stimulus 3; stimulus 2 and stimulus 4, and so on). Contiguous stimuli were not
presented together. This yielded 14 pairs of stimuli (seven groupings x two
orders), presented five times each (14 x 5 = 70 pairs in each comparison
task). The order of presentation of pairs was randomized for each participant.
The rating task consisted in the presentation of isolated stimuli, whose degree
of politeness participants had to rate on a five–point scale (“1” meant “not
very polite”, “3” was “adequate”, and “5” corresponded to “very polite”). In
Watts’s (2003) terms, “3” would correspond to politic behavior, i.e. behavior
that is adequate to the ongoing social interaction. There were nine stimuli
presented five times each (9 x 5 = 45 stimuli in each rating task). As in the
comparison tasks, stimuli were presented in a randomized order for each
speaker.
Before completing these tasks, participants were asked to sign a consent form
and fill in a questionnaire. They were asked to briefly define “polite” and
“impolite”. The test was administered so that participants first completed the
comparison task for a set of stimuli before undertaking the rating task with the
same set. The two other tasks were presented in the same order after a short
break. Some participants were exposed to falling stimuli first, while others
heard rising items first. The total duration of the testing session was
approximately 45 minutes.
In all the tasks, the same context was printed on screen and made clear to
participants. The context was the following: “Imagine you are taking a walk in
the street. A stranger approaches you and after greeting you, s/he asks: ‘do
you have the time?’”. In previous studies, context has been found to affect the
perception of politeness. Wichmann and Cauldwell (2003) asked students to
use their own labels to describe the attitudinal or affective meanings conveyed
by a series of sentences, first out of context and later in context. Participants’
16
perception changed drastically when sentences were embedded in their
context. Crucially, sentences in isolation were typically regarded as negative,
whereas, when inserted in a context, they received more positive ratings. In
another study by the same authors, it was observed that more polite
interpretations of utterances emerged in context; thus confirming Watts’s
(2003) claims that the context is crucial in order to assess the politeness of a
particular interaction.
2.5 Participants20 participants took part in this experiment (7 men, 13 women). All
participants were native speakers of Central Catalan living in Central–Catalan–
speaking areas, aged 18–44 (mean age: 30). None reported any known
hearing impairment.
2.6 Statistical analysesNon–parametric statistical tests were used to analyze the data. This choice
was motivated by the fact that the data sample was small (20 participants)
and that responses represented a rank which did not have any clear numerical
interpretation (a nominal scale was used).
17
3. RESULTS AND DISCUSSION
3.1 Rating TasksAs described in section 2, participants completed two rating tasks: one with
stimuli created from the rising question, and another one with stimuli created
from the falling question. They had to judge the degree of politeness of each
stimulus presented in isolation by assigning them a number, from 1 to 5.
Figure 5 presents the results of the two rating tasks. The graph reflects the
mean rating for each of the stimuli in the two continua. The solid line shows
the results obtained by the rising stimuli, whereas the dotted one represents
those obtained by the falling stimuli. The plot does not resemble the S–shaped
function usually expected in identification tasks because the contrast being
tested here is not categorical, but gradual.
First of all, note that the average ratings obtained by stimuli from both
continua (rising vs. falling) are very similar, suggesting that there may not be
differences in the degree of politeness conveyed by the two interrogative
contours available in Catalan and that both type of questions (rising and
falling) are equally suitable in terms of politeness for a context such as the one
presented in the experiment, i.e. a context of low cost for the hearer.
For the falling stimuli, note the flat line from stimulus –24 Hz to the
unmodified one, which indicates that participants failed to appreciate
differences in politeness among these stimuli. A slight gradual decrease in
perceived politeness is observed for the stimuli created by increasing the pitch
height of the boundary tone.
As for the line representing the ratings of the rising stimuli, it presents more
variation. The general tendency displayed is again a decline in politeness as
the pitch height of the final boundary tone rises. Notice that that seems to
contradict studies which suggest that rising or high pitch is associated to
secondary meanings such as subservience, politeness, etc.
18
The decline in politeness just described is the general pattern that emerges
when all participants are considered together, but there is variation across
participants. Some participants did indeed consider lower stimuli as more
polite, but some others showed the opposite tendency (politeness increased
as pitch height increased). Table 1 provides the mean rating and standard
deviation for each stimulus. As seen in figure 5, the difference between the
mean ratings for each of the stimuli is small, and the standard deviation is
generally high, suggesting variation across responses.
Rating of rising and falling stimuli
1
1,5
2
2,5
3
3,5
4
4,5
5
min24min18
min12min6
unmod
plus6
plus1
2
plus1
8
plus2
4
Stimuli
Rat
ingg
Rising Falling
Figure 5. Mean ratings for each of the stimuli in the rising and falling continua.
RISING STIMULI FALLING STIMULI
Stimulus Mean Standard deviation Mean Standard
deviationmin24 Hz 3.28 1.15 3.09 1.32min18 Hz 3.44 1.06 3.08 1.08min12 Hz 3.23 1.08 3.09 1.02min6 Hz 2.96 1.10 3.05 1.05unmodified 3.15 1.09 3.06 1.10plus6 Hz 2.90 1.15 2.84 1.09plus12 Hz 2.99 1.21 2.63 1.05plus18 Hz 2.84 1.24 2.58 1.17plus24 Hz 2.60 1.26 2.39 1.22
Table 1. Mean ratings and standard deviation for each of the stimuli in the continua.
19
Despite the small differences observable, the statistical analyses carried out
revealed some of them to be significant. Two multiple–sample Friedman tests
were performed for each set of stimuli and showed significant differences in
both continua (for rising stimuli: chi–square = 51.439; p < 0.000; for falling
stimuli: chi–square = 64.186; p < 0.000). To locate the differences along the
continuum, Wilcoxon two–sample tests were performed taking two contiguous
stimuli in the continuum. The difference between the extremes, and that
between the unmodified stimulus and each of the extremes was also
submitted to statistical analyses. Results are shown in Table 2. Columns two
and four present the p value obtained in the Wilcoxon test for rising and falling
stimuli respectively.
RISING STIMULI FALLING STIMULI
Pair p value Corrected p value
p value
Corrected p value
min24 – min 18 0.206 n.s. 0.924 n.s.min18 – min12 0.121 n.s. 0.680 n.s.min12 – min6 0.027 n.s. 0.708 n.s.min 6 – unmodified 0.152 n.s. 0.934 n.s.unmodified – +6 0.040 n.s. 0.058 n.s.plus6 – plus12 0.395 n.s. 0.076 n.s.plus12 – plus18 0.150 n.s. 0.507 n.s.plus18 – plus24 0.159 n.s. 0.057 n.s.min24 – plus24 0.000 * 0.003 *min24 – unmodified 0.334 n.s. 0.890 n.s.plus 24 – unmodified 0.000 * 0.000 *
Table 2. Results of the Wilcoxon tests carried out on the ratings of contiguous stimuli in each of the continua, the extremes and the mid–point, and the extremes. The new
alpha level is 0.0045.
Because multiple comparisons were tested using the same set of data,
Bonferroni correction was applied to the p values resulting from the Wilcoxon
tests to reduce falsely significant results. This type of correction consists in
dividing the alpha level (set at 0.05) by the number of comparisons carried out
with the same data set (0.05/11 = 0.0045). The new alpha level is 0.0045.
Columns three and five indicate the significant results after correction for
rising and falling stimuli respectively. Significant results are marked with an
asterisk (*). As shown in the table, the ratings received by stimuli at the ends
of the continua (those which had been modified by decreasing or increasing
the boundary tone by 24 Hz) were significantly different. Similarly, the
comparison between the ratings obtained by the unmodified stimulus and the
20
one with the highest boundary tone turned out to be clearly significant.
Differences in ratings for these two pairs were significant in both continua.
For both continua, significant differences are found between the unmodified
stimulus and the stimulus with the highest pitch values in the boundary tone,
and between the two extremes of the continua. Other comparisons did not
reach significance, neither for the falling stimuli nor for the rising ones. These
results confirmed what was ventured above: the higher the pitch in the final
boundary tone is, the less polite it is perceived to be. This is consistent with
some participants’ comments at the end of the perception test session
reporting that some of the highest stimuli sounded “irritating, impertinent”.
This effect is found to be clearly significant when we compare the extremes,
and the highest extreme and the mid–point of the continuum.
3.2 Comparison tasksThe experiment included two comparison tasks, in which stimuli were
presented in pairs (in the two possible orders: the lowest one followed by the
highest one, and vice versa). Participants had to determine which of the two
stimuli was perceived as the most polite and press a key accordingly. Stimuli
from the two different continua were presented separately in different tasks.
Figures 6 and 7 show the results of the comparison tasks separately for rising
and falling stimuli. The solid line represents the answers obtained by pairs
formed by the lowest stimulus followed by the highest one (LH order); the
dotted line represents the same stimuli in the reverse order (the highest
stimulus first, HL order). Responses were coded so that “0” means that the
lowest stimulus in the pair was considered the most polite, whereas “1”
means the opposite (i.e. the highest stimulus was regarded as the most polite).
21
Comparison of pairs of rising stimuli
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
min24_min12
min18_min6
min12_unmod
min6_plus6
unmod_plus1
2
plus6_p
lus18
plus12_p
lus24
Pairs
Low-High High-Low
Figure 6. Mean results of the comparison task for rising stimuli (0 = the lowest stimulus is more polite; 1 = the highest stimulus is more polite).
Comparison of pairs of falling stimuli
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
min24_min1
2
min18_min6
min12_un
mod
min6_plus6
unmod_plus1
2
plus6
_plus1
8
plus12_p
lus24
Pairs
Low-High High-Low
Figure 7. Mean results of the comparison task for falling stimuli (0 = the lowest stimulus is more polite; 1 = the highest stimulus is more polite).
22
The curves in both graphs indicate that the lowest stimulus in the pair was
generally preferred when the two stimuli were presented in LH order (that is,
when the first one was lower than the second). The HL order of presentation
(the highest one first) seemed to elicit more ambiguous responses. For the
rising stimuli, average answers are mostly situated around 0.5, showing that
there was no clear preference for one of the stimuli over the other when
presented in the HL order. For the falling stimuli, a similar tendency is
observed. However, note that in the HL order, the highest stimulus is preferred
in the first three pairs, whereas the lowest stimulus is preferred in the last
three pairs (i.e. pairs formed by combining stimulus whose final boundary tone
has been increased).
These differences across orders of presentation were submitted to statistical
analysis to verify whether participants were consistent in identifying one of the
two stimuli as more polite than the other regardless of the order of
presentation. Analyses were also carried out to determine whether there were
significant differences between the mean ratings of different pairs.
As done for the rating task results, an initial multiple–sample Friedman test
was performed: no significant differences were observed for pairs of rising
stimuli presented in the LH order (chi–square = 8.525, p = 0.202) nor in the
HL order (chi–square = 7.508, p = 0.276). Wilcoxon two–sample tests carried
out for each two contiguous pairs of stimuli were consistent with the first
analyses: no significant differences were observed.
Even if both continua patterned similarly in the rating task, differences emerge
here. Both Friedman tests for the comparison of pairs of falling stimuli in the
LH order and in the HL order returned significant differences (LH order: chi–
square = 31.083, p < 0.000; HL order: chi–square = 38.690, p < 0.000).
Contiguous pairs of stimuli were submitted to Wilcoxon tests. Significant
differences were found between two pairs: pairs four and five (min6–plus6
and unmod–plus12) when the stimuli were presented in the LH order and
pairs two and three (min18–min6 and min12–unmod) in the HL order.
However, the significance disappeared after applying Bonferroni correction
(corrected alpha level = 0.05/6 = 0.008; pair four–five: p = 0.016; pair two–
three: p = 0.046 ).
23
Because the nature of the contrast being tested in this experiment is gradual,
the inexistence of a category–shift boundary should come as no surprise. This
is precisely what is expected given the characteristics of the stimuli being
tested. Let’s turn to compare the responses obtained by the very same pair of
stimuli across the two orders. Order of presentation of stimuli was tested by
means of a two–sample (LH vs. HL) Wilcoxon analysis testing general
differences for all stimulus pairs. Figure 8 plots the means for pairs of rising
and falling stimuli presented in both orders (LH and HL). As can be seen,
participants were more inclined to evaluate the lowest stimuli in the pair as the
most polite when they were presented in LH order. In the HL order, average
responses are close to 0.5, suggesting less ability to classify which element
was the most polite. These differences are significant (pairs of rising stimuli: p
< 0.000; pairs of falling stimuli: p < 0.000). Thus, the order of presentation of
stimulus in a pair did affect participants’ judgments. Further Wilcoxon tests
were used to compare the pairings of the same stimuli in the two different
orders of presentation. Table 3 shows the results of the tests.
Presentation order effects
0,00
0,10
0,20
0,30
0,40
0,50
0,60
0,70
0,80
0,90
1,00
Order
Rat
ings
LH 0,37 0,31
HL 0,53 0,52
Rising Falling
Figure 8. Comparison of the mean responses of pairs of stimuli in the two orders of presentation (LH = the lowest one first; HL = the highest one first; 0 = the lowest stimuli
was perceived as the most polite).
24
RISING STIMULI FALLING STIMULI
Pair p value Corrected p value p value Corrected
p valuemin24_min12 0.115 n.s. 0.000 *min18_min6 0.036 n.s. 0.000 *min12_unmod 0.053 n.s. 0.052 n.s.min6_plus6 0.058 n.s. 0.034 n.s.unmod_plus12 0.423 n.s. 0.005 *plus6_plus18 0.005 * 0.004 *plus12_plus24 0.004 * 0.000 *
Table 3. Results of the Wilcoxon tests carried out on the ratings of the same pairs of stimuli in different orders. The new alpha level is 0.008.
Indeed, the means obtained by the same pair of stimuli when presented in the
different orders are significantly different for certain pairs. As was anticipated
by looking at the graph, the results obtained for each of the continua do not
pattern together as they did in the rating tasks. After Bonferroni correction was
applied, two pairs in the rising continuum received significantly different
ratings in the two orders, namely those formed by stimuli plus6 and plus18,
and plus12 and plus24. The falling continuum behaves differently. It presents
more differences: the pairs min24–min12, min18–min6, unmod–plus12,
plus6–plus18 and plus12–plus24 were rated differently depending on
whether they were heard in the HL vs. LH order.
Asymmetries derived from the order of presentation of the members in a pair
have been reported in the literature. Whether stimuli are presented in a rising
order (LH) or a falling order (HL) interferes with participants’ ability to
discriminate or compare the stimuli in the pair (see Vanrell, 2007). Apparently,
this is related to “declination”, a phenomenon consisting in the progressive fall
in fundamental frequency as the utterance progresses. Peaks that should be
equivalent may be realized at a lower pitch height toward the end of the
utterance (Gussenhoven, 2004). Several experiments have revealed that
listeners are able to compensate for this declination, or that they “use an
abstract declination line when judging the height of the contour”
(Gussenhoven, 2004:118; see also Gussenhoven and Rietveld, 1988;
Pierrehumbert, 1979; Vanrell, 2007).
This compensation or reference declination line can explain why results are
more ambiguous when the stimuli to be compared are presented in
25
descending order (i.e. when the second one is lower than the first, HL).
Compensation of later peaks may make it difficult for participants to perceive a
difference between the two. In the reverse order (LH), compensation effects
may emphasize the second peak so that it becomes more salient and more
distinguishable from the first one (Vanrell, 2007). Thus, asymmetries derived
by the order of presentation can stem from the effects of this compensation
mechanism.
To sum up, in the rising continuum, participants rated stimuli plus6, plus18,
plus12, and plus24 differently depending on the order in which they were
presented. The lower stimuli (plus6 and plus12) were considered more polite
when presented first, and the same is true for stimuli plus18 and plus24. In
the falling continuum, most of the pairs were rated differently in the two
orders. For pairs min24–min12 and min18–min6, again, the difference lay in
the fact that the first stimulus of the pair was rated as more polite than the
other, regardless of whether it was higher or lower than the other one.
However, in the pairs unmod–plus12, plus6–plus18, and plus12–plus24, the
lowest stimulus was always preferred, although this effect was more evident in
the LH order. It is precisely among stimuli with increased boundary tone that
the lowest one is more clearly preferred, regardless of order.
3.3 Discussion Following the trend opened by studies examining the effects of prosody on the
perception of politeness, the aim of this paper was to explore whether changes
in the pitch height (i.e. pitch scaling) of the final boundary tone in questions in
Central Catalan would result in different degrees of perceived politeness.
Whether a higher or lower pitch would lead to a higher degree of perceived
politeness was unclear, based on the different studies on the biological use of
pitch and the studies on impoliteness described in section 1.
Contrary to predictions based on the cross–species use of F0, politeness does
not seem to correlate with higher pitch in Central Catalan. Results of the two
rating tasks reveal the opposite effect: a tendency to evaluate higher final
boundary tones as less polite is observed and, most importantly, reaches
26
statistical significance in both continua. When comparing the two continua,
both unmodified stimuli receive higher politeness ratings than the stimulus
whose boundary tone was increased by 24 Hz. Note that no difference is
observed between unmodified stimuli and stimuli whose boundary tone was
decreased. Participants do not rate lower stimuli as being more or less polite.
It is only when there is an increase in pitch height that the degree of perceived
politeness diminishes. In the rating tasks, significant differences were
observed when comparing the ends of the continua as well as the unmodified
stimulus and the highest one for both continua. Comparisons of contiguous
stimuli did not reach significance at any point, although a gradual decline in
the degree of politeness as pitch of the final boundary tone increases is
observed.
It is interesting to observe that stimuli in both continua received similar
ratings. This may indicate that both types of questions (rising and falling) are
interchangeable in interactions that imply a low cost for the hearer. As
described above, Payà (2003) considered the falling contour as more polite
than the rising one. Payrató (2002) and Prieto (1998) also venture that the
choice of one contour over the other may be determined by the degree of
politeness that the speaker wants to convey. Further experiments targeting the
two types of questions are necessary in order to assess whether they are
indeed interchangeable and, if so, in what contexts.
The results of the comparison tasks support those obtained in the rating tasks.
Stimuli with lower final boundary tone are always preferred in the LH order,
and in some cases in the HL order. Interestingly, it is in cases where the
boundary tone was increased (in the falling interrogatives) that the lower one
is always preferred, no matter what the order of presentation was. Therefore,
the same effect is found in both the rating and the comparison tasks:
participants tend to consider stimuli with increased boundary tones as less
polite than the unmodified ones or those with a decreased boundary tone.
As described in the introduction, several studies have explored the biological
exploitation of pitch (Culpeper et al., 1993; Gussenhoven, 2002; Nilsenova,
2006; Ohala, 1984, 1996) and conclude that high or rising F0 is primarily
associated with being small. From this first association, secondary social
27
meanings stem: a high pitch is viewed as conveying submission,
subordination, and politeness, among other meanings. Apart from these
theoretically–based studies, experimental evidence for the relationship
between high pitch and higher degree of politeness has been found for
Spanish and Japanese. For Venezuelan Spanish, Álvarez and Blondet’s (2003)
production experiment revealed that participants uttered polite sentences with
a higher pitch height and higher F0 variability. In Orozco (2006), polite
utterances were seen to begin with a high boundary tone in Mexican Spanish,
whereas neutral utterances initiated with a lower boundary tone. A broader
use of pitch range (more peaks and valleys) was also evinced in that variety. In
Kumamoto Japanese, the presence of a late peak in the utterance or a final
rising boundary tone was perceived to be more polite than utterances with an
early peak or a falling final cadence, according to Maekawa (1999). In an
analysis of impolite speech, Culpeper et al. (2003) claim that a rising contour
may be more considered more polite because it keeps the conversation going,
whereas a falling contour tends to put an end to it. However, in the same study
they propose that extreme pitch may be disruptive to interlocutors, and thus be
regarded as impolite. Finally, Culpeper (2005) described a particular example
in which higher pitch average, among other prosodic cues, conveyed
impoliteness.
As shown above, higher pitch is not associated with higher degree of
politeness in this experiment. As a matter of fact, the opposite tendency is
revealed here: stimuli whose final boundary tone had been modified by
increasing the pitch were rated as less polite than the ones whose final
boundary tone had been decreased in pitch. Apparently, these results would
conflict with most of the studies described above. Considering the more
empirical studies, it is evident that their results are hardly comparable to the
ones exposed here. Both studies focusing on Spanish are production
experiments, and they actually make no claim about final boundary tone and
the role of its height. In these experiments, it is an initial high boundary tone
that is associated with politeness as well as F0 variability. These results are
not, then, directly comparable to the ones described here.
Maekawa’s (1999) study does focus on final cadence or boundary tone, but
the emphasis is on choice of F0 direction (rising vs. falling final boundary
28
tone). However, the author does say that the larger the prominence of the
rising boundary tone, the higher the degree of politeness perceived. These
results are in contradiction to the ones described here. This might be due to
differences in the pragmatic use of pitch height in Catalan and Kumamoto
Japanese. However, it should also be borne in mind that this variety of
Japanese has an idiosyncratic intonation that differs from that of Catalan (see
section 1). Finally, Culpeper (2005) noticed that higher pitch could also convey
impoliteness, but that was in combination with other prosodic cues. Again, it
was not focusing on pitch height of the final boundary tone, which makes a
comparison difficult.
To sum up, even if the results of this experiment seem to contradict those of
the studies described above, the experiment reported here aimed at isolating
the contribution of pitch height, whereas most of the other studies examined
the role of pitch in combination with other prosodic (or even lexical and
morphological) elements, such as duration, peak alignment, etc. The role of
pitch height might be different when exploited in conjunction with duration or
access to speaker’s face. Follow–up experiments may determine whether this
is really the case. It is important to note, however, that lower stimuli were not
perceived as more polite than the unmodified one either. There was simply no
difference between them. That is, the contrary tendency (i.e. the lower the
pitch, the higher the degree of perceived politeness) does not hold.
29
4. CONCLUSIONS
The goal of this experiment was to assess the contribution of pitch height of
the final boundary tone to the perception of politeness in two types of
interrogative contours in Central Catalan: the rising and the falling questions. It
was predicted that the degree of perceived politeness conveyed by the
question would be influenced by changes in pitch height.
To determine whether pitch scaling did have an effect on the perception of
politeness, 20 participants completed two rating and two comparison tasks.
The results obtained in the four tasks confirmed the initial hypothesis that
changes in pitch height interfere with the degree of politeness conveyed by
questions. Both types of tasks successfully captured the fact that participants
were sensitive to differences in pitch height and showed a preference for
unmodified or lower stimuli over stimuli with increased pitch in the boundary
tone. The higher the pitch in the final boundary tone, the lower the degree of
politeness it was thought to convey. Some participants felt that politeness
correlated positively with pitch. Further studies including duration and visual
access to the face of the speaker may shed more light on the role of prosody in
the perception of politeness.
Based on the results obtained in the rating and comparison tasks, it can be
preliminarily concluded that pitch height does play a role in the perception of
politeness. Stimuli with higher pitch are not considered to be as polite as
others with lower pitch. This seems to contradict a tendency for higher pitch to
correlate with politeness observed in some studies, but it may be a
consequence of examining the contribution of pitch in isolation or focusing on
different aspects of the utterance. It is possible that the contribution of pitch
height is different when combined with duration or other visual and acoustic
cues, and it is also possible that languages exploit pitch differently in order to
convey paralinguistic meanings. As other studies have demonstrated before, a
description of politeness (whether of an abstract, theoretical, and universal
30
nature or not) may benefit from the incorporation of prosody as a decisive
element in the encoding of politeness.
31
5. REFERENCES
Álvarez, Alejandra and Blondet, María Alejandra. 2003. Cortesía y prosodia: un estudio de la frase cortés en el español de Mérida (Venezuela). In: Herrera Z., E. and Martín Butragueño, P. (Eds.), La tonía: Dimensiones fonéticas y fonológicas. México D. F.: El Colegio de México, 319–331.
Boersma, Paul and Weenink, David. 2005. Praat: doing phonetics by computer (Version 4.5.02) [Computer program], University of Amsterdam. Available online at http://www.praat.org/.
Bonet, Eulàlia. 1984. Aproximació a l’entonació del català. BA thesis, Universitat Autònoma de Barcelona.
Brown, Penelope and Levinson, Stephen C. 1987. Politeness. Some universals in language use. Cambridge: Cambridge University Press, 1978.
Chen, Aoju. 2005. Universal and language–specific perception of paralinguistic intonational meaning. PhD dissertation. Utrecht: LOT.
Culpeper, Jonathan. 2005. Impoliteness and entertainment in the television quiz show: The Weakest Link. Journal of Politeness Research 1, 35–72.
Culpeper, Jonathan; Bousfield, Derek; and Wichmann, Anne. 2003. Impoliteness revisited: with special reference to dynamic and prosodic aspects. Journal of Pragmatics 35, 1545–1579.
Escandell Vidal, M. Victoria. 1988. La interrogación en español: Semántica y Pragmática. PhD dissertation. Madrid: Editorial de la Universidad Complutense.
Estrella–Santos, Ana. 2007. Politeness and prosody: Interaction of prosody in the requests with “dar” as a benefactive in Ecuadorian Andean Spanish. Paper presented at the Phonetics and Phonology in Iberia 2007 Conference (PaPI), Braga (Portugal), June, 24–27.
Forster, Jonathan. 2007. DMDX Display Software. [Computer program]. Available online at: http://www.u.arizona.edu/~kforster/dmdx/dmdx.htm.
Gussenhoven, Carlos. 2002. Intonation and interpretation: Phonetics and Phonology. Speech Prosody 2002: Proceedings of the First International Conference on Speech Prosody. Aix–en–Provence, ProSig and Université de Provence Laboratoire Parole et Language, 47–57.
Gussenhoven, Carlos. 2004. The phonology of tone and intonation. Cambridge: Cambridge University Press.
Gussenhoven, Carlos and Rietveld, Antoni. 1988. Fundamental frequency declination in Dutch: Testing three hypotheses. Journal of Phonetics 16, 355–369.
Hualde, José Ignacio. 2005. The sounds of Spanish. Cambridge: Cambridge University Press.
Ide, Sachiko. 1982. Japanese sociolinguistics: Politeness and women’s language. Lingua 57, 23–43.
Maekawa, Kikuo. 1999. Contributions of lexical and prosodic factors to the perception of politeness. In: Proceedings of ICPhS99 [14th International Congress of Phonetic Sciences, San Francisco]. Berkeley: University of
32
California. Available at http://www2.kokken.go.jp/~kikuo/public/ICPhS1999.pdf.
Nilsenova, Marie. 2006. Rises and Falls. Studies in the semantics and pragmatics of intonation. PhD dissertation. University of Amsterdam.
Ofuka, Etsuko; McKeown, Denis J.; Waterman, Mitch G.; and Roach, Peter J. 2000. Prosodic cues for rated politeness in Japanese speech. Speech Communication 32, 199–217.
Ohala, John J. 1984. An ethological perspective on common cross–language utilization of F0 of voice. Phonetica 41, 1–16.
Ohala, John. J. 1996. Ethological theory and the expression of emotion in the voice. In: Proceedings of ICSLP 96 [4th International Conference on Spoken Language Processing, Philadelphia]. Wilmington: University of Delaware, vol. 3, 1812–1815.
Orozco, Leonor. 2006. Peticiones corteses y factores prosódicos. Paper presented at the Fonología Instrumental: patrones fónicos y variación lingüística Conference, Mexico, October 23–27. Available at http://lef.colmex.mx/Sociolinguistica/Cambio%20y%20variacion/Orozco.pdf.
Payà, Marta. 2003. Politeness strategies in spoken Catalan: when prosodic and non prosodic elements don’t match. Paper presented at the 6th NWCL International Conference: Prosody and Pragmatics, Preston (United Kingdom), November 15.
Payrató, Lluís. 2002. L’enunciació i la modalitat oracional. In: Solà, J.; Lloret, M–R.; Mascaró, J.; and Pérez Saldanya, M. (Eds.): Gramàtica del català contemporani. Barcelona: Editorial Empúries, 1151–1222.
Pierrehumbert, Janet. 1979. The perception of fundamental frequency declination. Journal of the Acoustic Society of America 66, 363–369.
Prieto, Pilar. 2001. Notes sobre l’entonació dialectal del català: Les oracions interrogatives absolutes. In: Bover, A.; Lloret, M–R.; and Vidal–Tibbits, M. (Eds.): Actes del Novè Col·loqui de la North American Catalan Society. Barcelona: Publicacions de l’Abadia de Montserrat, 347–377.
Prieto, Pilar. 2002a. Entonació. In: Solà, J.; Lloret, M–R.; Mascaró, J.; and Pérez Saldanya, M. (Eds.), Gramàtica del català contemporani. Barcelona: Editorial Empúries, 393–462.
Prieto, Pilar. 2002b. Entonació. Models, teoria, mètodes. Barcelona: Editorial Ariel.
Prieto, Pilar. In press. The intonational phonology of Catalan. In: Jun, Sun–Ah (Ed.), Prosodic Typology 2. Oxford: Oxford University Press.
Prieto, Pilar and Rigau, Gemma. 2007. The Syntax–Prosody Interface: Catalan interrogative sentences headed by que. Journal of Portuguese Linguistic 6:2, 29–59.
Prieto, Pilar; Aguilar, Lourdes; Mascaró, Ignasi; Torres–Tamarit, Francesc Josep; and Vanrell, Maria del Mar. 2009. L’etiquetatge prosòdic Cat_ToBI. Estudios de Fonética Experimental XVIII: 287–309.
Vanrell, Maria del Mar. 2007. The phonological role of tonal scaling in Majorcan Catalan interrogatives. MA thesis. Barcelona: Universitat Autònoma de Barcelona.
Watts, Richard J. 2003. Politeness. Cambridge: Cambridge University Press. Wichmann, Anne and Cauldwell, Richard. 2003. Wh–questions and attitude:
the effect of context. In: Wilson, A., Rayson, P., and McEnery, T. (Eds.), Corpus Linguistics by the Lune: A Festschrift for Geoffrey Leech. Peter Lang: Frankfurt am Main, 291–305.
33